Characterisation of alternative exon usage between and within adipose and skeletal muscle

Introduction: Current molecular biology methodologies interrogating Alternative Exon Usage (AEU) events such as alternative splicing, promoters and polyadenylation are still limited. Further, these co-transcriptional mechanisms are understudied in metabolic scenarios. Alternative transcripts produced from metabolically relevant genes should produce an alternative protein product with a different function from the canonical protein. Thus, we aim to develop a novel method that can detect AEU events accurately on a high?throughput basis. In parallel, we also aim to develop a method that measures transcript ratios on a single gene basis. Our methodologies were validated within/between adipose and muscle tissue due to their significant roles in energy metabolism but also because they both originate from mesenchymal stem cells (MSC). Methods: Large-scale exon expression was assessed using microarrays that measure each individual exon within the blood (n=13), muscle (n=14) and adipose tissue (n=14) from the FITFATTWIN cohort. These values were used to develop and validate our method: integrated Gene and Exon Model of Splicing (iGEMS). From a subset these validated genes we also demonstrate our method to measure transcript ratios which use real-time quantitative PCR (rt-qPCR) in combination with exon junction specific primers. We also employed a motif enrichment method called Analysis of Motif Enrichment (AME) to identify RNA-binding proteins (RBP) that are responsible for muscle and adipose tissue specific AEU events. Whilst comparing muscle and adipose is important developmentally, we utilised iGEMS on murine adipocytes undergoing ‘browning’ (n=8) when compared to white adipocytes (n=9) and also investigated ‘browning’ markers in human white adipose tissue from the IDEAL study (n=79). Finally, we applied iGEMS on skeletal muscle from insulin resistant (IR) individuals (n=14) and compared to healthy matched controls (n=14) from the METAPREDICT study. Results: IGEMS identified >4000 genes undergoing an AEU event between blood, adipose and muscle, from three pair-wise comparisons (FDR<5%). A subset 22 AEU events from the muscle vs adipose comparison that fairly represent this group were validated with RT-qPCR using exon specific primers and overall we achieved a 95% success rate. Next, we applied our transcript ratio method on validated genes and we were successfully able to measure these ratios for WDR7, CAPZB, CSDE1, SRSF5, TTC17, STK40, STAU1 and HNRNPM. In order to determine RBP that regulate these transcript ratios, we applied AME on each exon with flanking sequences from muscle and adipose tissue. We also integrated RBP gene expression. This resulted in 5 adipose tissue specific RBP such as RBMS3 and 12 muscle specific RBP such as RBM24 that regulate AEU within their respective tissue. We also show how transcript ratios can be used on clinical samples, as we demonstrate the transcript ratio of the calcium associated gene: RYR1 correlates with maximal voluntary contraction (p=0.048, R2 = 0.22). We also applied iGEMS on adipocytes undergoing ‘browning’ and we identified 555 genes (FDR<1%) undergoing an AEU event. From these genes, ~ 90% were successfully validated such as Agpat1 and Clstn3. Using this same dataset, we explored browning markers and found 60% agreement when compared to another study. The markers in agreement were then used to assess the impact of exercise and weight-loss on the ‘browning’ status of the white subcutaneous white adipose tissue. From potential phenotypic markers, we found that only UCP1 (p=0.006, R2 =0.09) and NATL8 (p=0.03, R2 =0.11) significantly correlated with weight?loss but was in the wrong direction if ‘browning’ was to occur. Using non-bias methods, we identified 181 genes that significantly correlated with weight-loss and these genes were associated with adipogenesis via C/EBPα signaling (z score = 2, P = 6.6 X 10-7 ). Lastly, iGEMS was applied on skeletal muscle form two groups with significantly different insulin AUC120 and identified >400 genes that underwent an AEU events. A select number of biologically relevant genes were validated using the whole METAPREDICT cohort (n=119). We found that only UQCRC2, TROVE2 and PKM successfully validated, and the RBP: ZFP36 is a potential regulator controlling these events. Discussion: Our new method: iGEMS was applied between blood, muscle and adipose tissue, and we were able to identify ~5-10 more AEU events than a recent RNA-seq study. Further, we identified a variety AEU events and were genuine as they were successfully validated by RT-qPCR. Thus, iGEMS demonstrates itself as a valuable tool to identify AEU events and was used in subsequent chapters. In addition to iGEMS, we developed a RT-qPCR method to accurately measure transcript ratios which are often overlooked in many AEU studies and we demonstrate its utility. When iGEMS was applied between white and brown adipocytes, this was the first time AEU was shown to occur and not just typical gene expression. These events were genuine as were successfully validated with RT-qPCR. Using a stringent set of browning markers, we found no change in subcutaneous white adipose tissue in response to combined exercise and calorie restriction. This rules out the contribution of myokines towards the ‘browning’ process but we have shown white adipose tissue responding by undergoing adipogenesis. The final data-set that iGEMS was applied to was between control and individuals with glucose intolerance. Gene expression differences within skeletal muscle from glucose intolerant individuals are limited but we demonstrate clear examples of AEU differences, and even more striking are exons expressed in an ‘on-and-off’ pattern. In conclusion, our novel method iGEMS revealed novel AEU events which provide motivation for the continual development of new biological tools.